E. Michailidi, Investigation of flood simulation for design in ephemeral basins - Application to Sarantapotamos basin, Diploma thesis, 140 pages, Department of Water Resources and Environmental Engineering – National Technical University of Athens, December 2013.
The typical hydrological procedure used for design in Greece includes empirical methods that were experimentally tested for specific types of basins abroad. Consequently, while implementing those methods in Greek basins the results can be unrealistic or misleading. This procedure was tested in a specific part of Sarantapotamos basin with measured flow and rain data and results were later tested and analyzed. Specific interest was given in the efficiency of the SCS method using typical as well as calibrated parameters in this type of limestone basins with ephemeral flow that favours deep percolation. The method couldn’t simulate the observed flows mainly due to the existence of intense interflow that exceeded in duration the concentration time and also the method’s inherent inability to take into consideration the change in soil moisture that occurs in an episode. After adding and optimizing an additional parameter used in the parametric SUH the optimized SCS method produced very satisfying results. Next, a hydrological analogues model was developed using six parameters and four initial conditions. The model was calibrated based on the available rain and flow data and the simulated flows were calculated. These flows were later compared with those obtained from the optimized SCS method and the results proved to be similar. In addition, synthetic rain distributions were produced using a stochastic procedure as well as an empirical procedure (alternating block method). The synthetic rain distributions were transformed into flow with the implementation of the hydraulic analogues model and a statistical analysis of the hydrographs was carried out. The statistical analysis of the peak flows showed an underestimation of the peak flow when using the alternating block method and therefore leads to the conclusion that stochastic methods used for rainfall distribution are safer to use for the hydrological design. Results from this thesis were presented at the 5th EGU Leonardo Conference.